The AI sovereignty movement is gaining traction as individual countries seek a “Third Way” alternative to the dominance of Chinese and American AI stacks. The goal is to ensure control of local data and governance but reaching that finish line may depend on access to key AI infrastructure like supercomputers.
The appeal of sovereign AI may have been underestimated. John Roese, global chief technology officer and chief AI officer for Dell Technologies, told an APAC regional gathering this week that the sovereign AI sector would be larger than many current forecasts suggested while noting that the success of these initiatives would depend on underlying infrastructure and data platforms.
Roese may have been reacting to a just-hatched Taiwanese sovereign AI initiative. Taiwan, a country that exists in the shadow of a hostile China, said the backbone of its plan to become an “AI Island” is a new “Nano 4” supercomputer equipped with 1,760 NVIDIA H200 chips and 144 of the company’s Blackwell chips.
“This proves to the world that Taiwan not only has advanced chip manufacturing but also possesses world-class strength in system integration and high-performance computing,” said Lai Ching-Te, president of Taiwan.
The importance of having a supercomputer as a backbone for AI sovereignty already is evident in Denmark, a country that leads Europe with a 28 percent AI adoption rate, according to data from the World Economic Forum.
Denmark is very keen to strike a balance between innovation and regulation. The Scandinavian country is expected to pass a law in 2026 that would grant Danish citizens copyright over their own images, a move meant to counter the proliferation of AI deepfakes and put social media platforms at risk from steep fines. Denmark also is banning the use of social media for children under the age of 16, a move that follows a similar restriction in Australia.
A “Third Way” conceptually looms large for Denmark as the country’s military intelligence service has classified the U.S. as a security threat, citing the imposition of high tariffs and a seeming willingness to use military force even against allies, a reference to President Trump’s expressed desire to control Greenland.
Denmark’s unease underlines broader worries about swift shifts in U.S. policies or export controls that are galvanizing an AI sovereignty movement that wants innovation, not dependency. The Chinese alternative presents issues like surveillance, data access, and ideological conformity.
Insight into what may prove to be a model for “Third Way” AI sovereignty came from Nadia Carlstein, chief executive officer of the Danish Centre for AI (DCAI) at the recent AI Summit in New York City. Carlstein emphasized that sovereign AI ensures control over data and compliance with national regulations.
A key element in Denmark’s AI initiative is a supercomputer called Gefion, an investment now beginning to bear fruit. Gefion made its presence known in November with a joint AI project of DCAI and Siemens Gamesa to fine-tune wind turbine manufacturing and operations.
“Using Gefion’s state-of-the-art AI capabilities and AI-supported product development will allow us to further realize the potential of wind power,” said Vinod Philip, a Siemens executive. “Societies will greatly benefit from this.”
Gefion uses NVIDIA GPU clusters to support a wide range of activities ranging from traditional high performance computing to AI-specific tasks like model training. DCAI encourages experimentation by allowing both open source and proprietary software and knowledge-sharing among users. The idea is to make AI accessible to both sophisticated users and newcomers to the field.
Carlstein cited Gefion benefits that have led to much improved forecasts by the Danish Meteorological Institute. Other examples included companies that have leveraged AI for quick commercial transactions and drug discovery as well as improved diagnostic models of patients in health care. The development of a sovereign AI infrastructure means Denmark has a secure environment for AI development across many industries, said Carlstein.
Meanwhile, the need to provide AI sovereignty infrastructure is creating opportunities. Quest Software, for example this month opened its Centre for Advanced AI Architecture in Cork, Ireland that will create 100 jobs. The move dovetails with European Commission’s AI Continent Action Plan to attract more AI talent.
AI data sovereignty and the infrastructure to support it is a strategic concern. When it comes to AI, more and more countries want to control their own destiny.
